Okay, everyone, I think we can get started. My name is Steve Cantor. I'm the Vice President of Investor Relations for Entegris, and it's my pleasure to welcome everyone to the 2018 Entegris Analyst Meeting, both the people in the room today, here at our headquarters in manufacturing and R&D facility in chilly but very sunny Billerica, Massachusetts, as well as those of you on the webcast. This meeting is being webcast live and will also be available on demand. The webcast, as well as an electronic copy of the slides, are available on our website. For the people in the room, I have a few housekeeping items to review before we can begin. First, we'd like to ask that you hold your questions until all the presentations are completed. At approximately 12:00 P.M., we're going to take a 10-minute break.
For those in the room, you can grab a boxed lunch, which will be positioned right behind the curtain behind you. We expect to conclude the formal part of the meeting at 1:30 P.M. For those here today, we are offering a tour of our manufacturing area, including the recently acquired Trinzik line of filters, as well as our analytical labs. Those who are interested should meet in the back of the room following the Q&A. You'll notice that there are surveys in front of you. We really appreciate you completing those. Your feedback is extremely helpful to us to make these meetings even better. Restrooms are in the back to the left. You'll see signs. Also, in the unlikely event of an emergency, the nearest exit is the door you entered the building. That's back to the left, down the stairs, and out the door.
I'd like to remind everyone, both here and on the webcast, that we will be making forward-looking statements today, and we encourage you to read our filings carefully to understand the risks and uncertainties related to those statements. We have a great agenda today, which will cover many of the reasons why we're excited about Entegris' future. Our speakers today will include Bertrand Loy, President and Chief Executive Officer; Wenga Yang, Vice President of Strategic Marketing; Todd Edlund, Chief Operating Officer; Jim O'Neill, Chief Technology Officer; and Greg Graves, Chief Financial Officer. I'd like to introduce a few of the other members of the management team that are here today. First, you can raise your hand. Sue Lee, SVP, Chief Counsel. Sue Rice, SVP of Human Resources. Sue's in the back. Greg Marshall, SVP of Quality, EH&S, and Business Support. Corey Rucci, SVP of Business Development.
Stuart Tison, SVP, who heads our Specialty Chemicals and Engineered Materials. Maybe he's not here. Then Clint Haris, who is SVP and runs our Microcontamination Control business. Clint will be leading us on the tour following the formal part of the meeting. Also, I'd like to welcome Dr. Azita Saleki-Gerhardt, who's our newest member of the Board of Directors, and she's in the back. Before turning it to Bertrand, I do want to, again, remind you to hold your questions until the formal presentations are completed. With that, Bertrand.
Good morning. Thank you, Steve. Thank you again, all of you, for joining this annual Entegris Analyst Day, whether you do this in person here or over the webcast. As the CEO of Entegris, I'll tell you that I'm very proud of what we have accomplished in the recent years as we outpace the industry, as we outpace our peers. As a management team, what we want to tell you today is that no matter how good we feel about the recent performance of Entegris, we are yet more excited about what lies ahead. The objective for us today is simply to share with you the reasons for that optimism. Going to next slide, I will start with what used to be a prediction of mine, it's probably now more of a fact. The fourth industrial revolution is here.
Some would call it the digital revolution. I prefer to call it the data revolution. The fourth industrial revolution, indeed, in my opinion, is about creating intelligence, extracting value out of the data that will be collected, the data that will be stored, processed, indexed, correlated. In other words, it's really about moving data from the very edge of the network back to servers to repurpose it, to move the data back to the users in the form of new insight. The users can be humans, the users can be machines. That new insight that will be created is what will transform our healthcare, will transform our cities, will ultimately transform so many aspects of our lives. You would say, "Okay, well, nothing really new here. What does that mean for the industry?
What does that mean for Entegris?" Let me talk about all of those and address all of those questions. First, the explosion of data will drive greater innovation, and that, in turn, will create the conditions for a proliferation of chips. Those examples here on the screen are actually presaging. We estimate that by 2025, 75 billion devices will be connected to the internet, and all of those devices will be equipped with a multitude of sensors, collecting and generating huge amounts of data. Google estimates that an autonomous car will generate about four terabytes of data a day. Take a couple of those cars and combined, they will generate an amount of data equivalent to the printed collection of the U.S. Library of Congress. Fast-forward 2025, Seagate estimates that by 2025, on an annual basis, we will collectively generate about 160 zettabytes of data.
That's a lot of gigabytes. That's actually a lot of Libraries of Congress, if you think about it. Beyond those examples, which at some level can be helpful, but on another level can be constraining, what is really important for all of us to remember is that improvements in the technology efficiency will drive new usage. Those new usage ultimately will translate into greater rate of consumption of semiconductors. The takeaway for you here is that the digital revolution will create conditions for the semiconductor industry to grow at a level of about 4%-5% a year for the next three years. Here, we're talking about chip volumes. As you know, the volumes of chips manufactured every year is the primary driver for the Entegris business.
This is all good and well, but for that digital revolution to actually become a reality, we will need a better infrastructure. What I mean by that is we will need sensors that are more reliable. We will need storage solutions that are more effective. We will need a network with better latency, and we will need faster computing power to essentially create that insight that I was talking about in a near-instantaneous way. Let me tell you, we are not quite there yet. Jim O'Neill, our CTO, will tell you that when you think about the next level of performance that semiconductors will need to offer, well, miniaturization alone, Moore's Law alone will not be the answer.
Instead, the semiconductor industry will turn to new materials with better electrical properties and will also turn to new structures, new architectures for the chip to enable, again, that greater level of performance. That's great news for Entegris, because if you think about the materials that would be required to make those chips, if you think about the chemistries that would be required to make those chips, they would be subject to very, very tight specs in terms of quality, in terms of purity, in terms of stability. These specs will really push the limits of what we know how to do in this ecosystem. It will require the industry ecosystem to collaborate to solve those emerging process challenges.
When you think about Entegris, and if you think about the three divisions comprising Entegris, SCEM, Microcontamination, Advanced Materials Handling solutions, we will be a critical enabler for the process solutions of tomorrow. Todd Edlund, our COO, will describe how those various technology platforms will be very complementary and will be very critical for the semiconductor ecosystem to enable those new materials, to enable those new structures. The punchline here and the takeaway for you is that what we do at Entegris is becoming increasingly important for the semiconductor technology roadmap. As a result of that, we expect our served market to grow faster than wafer starts. In the previous slide, I was talking about collaboration, and I grant you, everybody talks about collaboration these days. At Entegris, we have placed collaboration at the center of our business model for the last 10 years.
It's really truly part of our DNA today as a company. As a result, I would argue that we are probably much better at collaborating than most companies. If you think about our value propositions, it's really centered around three major elements. The first one is superior technology. Over the last five years, we have increased very significantly the amount we're spending every year on R&D. More importantly, we have done a much better job at making sure that we are very closely aligning our technology roadmap to the emerging process challenges of the industry. Jim O'Neill, our CTO, will explain to you how, as a result, our portfolio has evolved over time. The second very important part of our value proposition is the global infrastructure. If you want to collaborate, you just have to make it very easy for the customers to do that with you.
We have invested in tech centers, we have invested in metrology capabilities, we invested in talent in all of the major markets in the semiconductor industry to make it possible for customers to collaborate in the same language, in the same time zone. When you think about the teams interacting every day with our customers, you would find people that are very smart, and Jim will talk about that, but also people that are really team players. You will find people that are really fast at developing new technology, but also very disciplined and rigorous and adhering to very prescribed development protocols. In other words, our customers love to work with our teams. Jim should take great pride of that, and he will talk to you about all of the investments we've made on that front.
Last but not least, operational excellence has been a major area of focus for us. In the end, what our customers want is unique technology that we can actually, if they like the technology, that we can move very quickly from the pilot stage to high volume manufacturing. They want that to be done in no time and in very tight process windows. I think we're getting pretty good at that. That has required very significant investments over the last decade, investments in statistical process control, in talent, again, and increasingly, over the last few years, a greater focus in managing our own supply chain so we can actually minimize variability in the input materials, so we can actually enable early fingerprinting of the input materials.
In other words, again, the takeaway for you here is that we have established Entegris as the preferred partner for most of the technology leaders in this industry, and that, over time, will translate into market share gains. This slide will allow me to summarize the previous three slides. Remember, first big trend is this digital revolution will actually create very favorable lasting conditions for growth in the semiconductor industry. We expect, as a result, very steady growth, in terms of wafer starts. Second aspect is what we do at Entegris is increasingly important for this industry. We talked about the importance of materials, we talked about the importance of purity, and that will translate into a fast expansion of our served markets.
The third point is, I talked about the unique value proposition, and I talked about the quality of our execution, and that will translate into market share gains. If you blend all of those considerations together, you have actually what is our long-term growth objective of about 5%-8% of annual CAGR. Now, before I turn the page, there are a few other considerations at the bottom of the slide that I would like to touch on, because those are very unique attributes that make Entegris stand out among other investment options that you may have in this space. First, we have a diverse customer base, and Todd Edlund will talk to you about why the very nature of what we do, why our position in the ecosystem is enabling us to have that diversity in our customer base.
The other important piece is that most of what we do are consumable products that are used in the daily production cycles of the semiconductor fabs. Our process solutions are very sticky. All of that translates into very long tails of recurring businesses and recurring revenues. Todd Edlund will actually provide a little bit more details around all of that. The other aspect is really the fact that we will often refer to the breadth of our portfolio. What it means for you as investors is that at every turn in the industry technology roadmap, we've been able to find opportunities for some parts of our portfolio. Jim O'Neill will explain to you how we are managing innovation at Entegris in a broad way, and how, as a result, we are nearly agnostic to any major technology shifts in the industry.
Great organic growth prospects for Entegris. Let me be clear, the focus of this management team is not limited to top-line growth. As we grow the top line organically, we intend to maintain what we consider to be a very attractive flow-through of 40%. What it means is that ultimately, we intend to grow our bottom line, we intend to grow our cash flow at a faster rate than our top line. Todd Edlund will provide actually some elements of the strategy that will allow us to control our cost structure to provide that leverage. Greg Graves, our CFO, will actually drive you through a fairly comprehensive walk forward of what it means for the years to come.
Another big area of focus, obviously, is the capital allocation strategy. Entegris is a business that has been generating a lot of free cash flow, and we certainly expect that to continue going forward. The question is, what will we be doing with the cash we are generating? That slide actually addresses that question. As I mentioned, we believe that our core business has a lot of potential. No surprise, our number one priority will be to reinvest in the business, to shore up the value proposition I was describing earlier. In other words, make sure that we spend adequately in R&D, make sure that we continue to invest in the global infrastructure, advanced metrology, and of course, continue to add to our manufacturing capabilities and manufacturing capacity. Next would come acquisitions.
We believe that we have proven to be a very effective acquirer, and we expect to be able to be more active on that front going forward. At the bottom of the slide, you can see some other elements of our capital allocation strategy. Dividends. We announced actually a quarterly dividend in Q4 of last year, an ongoing share repurchase program, and then the continuation of our term loan repayment, something that we started right after the ATMI acquisition in 2014. Greg will provide a lot more details around all of those various elements. When it comes to acquisitions, you can expect us to stick to a series of very clear guidelines against which we will be assessing any new acquisition candidates.
From a technology standpoint, we would love to continue to strengthen and expand our capabilities in advanced materials, in separation technology, in sensing and control, and advanced packaging. From a market standpoint, expect us to continue to focus on our semiconductor market, but in some cases, we may choose to venture in adjacent markets if we can find acquisition that could accelerate internal development work and could provide opportunities for the broader Entegris portfolio. Finally, you can always expect us to adhere to very strict financial guidelines. Financial guidelines that we review routinely with our board, and financial guidelines that would include considerations around minimum levels of accretion year two after an acquisition, minimum levels of ROIC three years after a deal.
Another way to actually talk about acquisition is just to state that we will be focusing on acquisitions that create long-term value for our customers, long-term value for our shareholders. If you look back at what we've done over the last three, four years, I think that the recent deals actually are perfect proof points of the focus and the discipline that I was actually describing. Let me talk a little bit about that. With ATMI, we got access to a whole new scale, and we got access to a very complementary platform. I'll give you a few examples of that. First, the scale obviously was a big enabler for the EBITDA expansion in 2015. Scale was actually as important in our ability to attract talent as a company, attract talent in Taiwan, attract talent in Korea, and even in the U.S. here.
I talk about complementarity, that was true from an infrastructure standpoint as well. Think about it this way. ATMI brought to us a tech center, a manufacturing footprint, better customer access in Korea. I will tell you that the success we recorded in Korea last year is in great part related to some of those benefits that we got from the ATMI acquisition. That, frankly, benefited every single product line across our portfolio. Again, if you think about ATMI, Jim O'Neill, our CTO, will talk about the new value proposition that we've been able to create for our customers. He will give you examples of co-optimization of filtration and packaging for our cleaning solutions. He will talk about system solutions enabling usage and dispense of advanced deposition materials on the wafer. I will leave that to Jim. Moving on to Trinzik.
Slightly different scale, more of a tuck-in, more of a product line expansion, given us access to larger filters focused on bulk chemical applications in the sub-fab. Something that traditionally we would not have really focused very much on. A very nice extension. Small, but we expect this transaction to contribute $0.01 to $0.02 in 2018 and 2019. Finally, PSS. Great technology, great team, very solid platform that I think will blossom as it gets access to our global distribution network and as PSS can get access to our supply chain and manufacturing best practices. It's a business that has a strong foot in semiconductor applications with sensing and control capabilities around the CMP modules, also provides very nice opportunities in life sciences as well. This slide actually concludes my presentation.
When you listen to the next presenters, I hope we can press upon you that Entegris is a very unique platform. Entegris is, in fact, a very unique investment option for you. We have very exciting organic growth potential. We have a very strong balance sheet, a balance sheet that we intend to put to work to generate EPS in excess of $2.50 by 2020. With that, I will turn to Greg.
Okay. Thank you very much, Bertrand. While we transition to the next speaker, just sort of key highlights of Bertrand's presentation. At a level, he's told you the whole story. We've got a great market driven by the digital revolution, that digital revolution means more chips, more complex chips, all of which benefit us. We're a key enabler to the industry. We're sort of on-trend in terms of what we're doing, in terms of material solutions, contamination control. What does that mean to me? What does it mean to an investor? We should win no matter what. I mean, just the fact that there's more volumes, more complexity, what we do really matters. The other thing that he didn't mention so much in words, but underlying thesis, is the discipline that we have as a management team.
I think if you look at our historical execution and as you look forward, we have been a team that sort of, we say what we're going to do, then we go out and do it. We don't swing for the fences, but we've delivered on our commitments repeatedly over the last three to five years. With that, one housekeeping issue, Steve forgot to mention Bruce Beckman, who is my key lieutenant on the finance team. He's our Senior VP of FP&A. Played a big role in putting this together and plays a big role in sort of driving our operations with Todd day to day. With that, I'm going to turn it over to Wenga, who is our VP of Marketing Strategy. Wenga's been with us about six years.
Came to us via AMD, Tokyo Electron, short stint on the sell side at Citibank, don't hold that against him, then came to us, like I said, about six years ago.
Thank you, Greg. Good morning, everyone. Bertrand has give a very good overview of what the company is facing and what's our opportunities. I would dive a little bit deeper into what's the market trend or what does that mean to Entegris. If you look at the overall digital era, digital age that everyone's talking about, what does that mean specifically to Entegris? What we see is with this digital evolution, there are multiple industry drivers that's helping the semiconductor industry. All those drivers simply means more semiconductor chips needed for the whole society to enable different applications and different new growth areas. Those chip demands drive the fundamental matrix that we follow, which is the MSI or million square inch wafers annually. Those are the wafer stocks that Entegris revenues have very closely correlated. Because we need more chips, we will stock more wafers.
Hence the revenue fundamentals of Entegris is right there. There are two additions to the more wafer stocks that's beneficiary to Entegris. We will discuss in a little bit more detail. The first one is that for every single wafer stock, actually, we are increasingly using more materials. As we know, Entegris is a very specialty materials company. More material usage added the growth elements to Entegris. Also, the biggest challenge our industry is facing as we continue to progress on the technology is how to control the contaminations. Contamination control become one of the biggest challenge of the industry, and it's one of the biggest opportunities for Entegris. If you look at this whole flow, more drivers, more wafers, more materials per wafers, and all the materials has to control contamination. This basically set up the overall market trend for the company to grow.
Diving a little bit more deeper, and there are a lot of buzzwords here. What I want to say is, if we look at what's driving different type of semiconductor chips for logic and memory, and we all hear about Samsung, Intel, TSMC, they're talking about there are multiple new applications kicking in. In the past, it's PC, it's smartphone. Today, everybody's talking about artificial intelligence, the 5G network, data center, big data, VR, ARs, and also cryptocurrencies. Many of those new applications are driving the progress of logic and memory. There's an area that people normally don't pay a lot of attention, which we call the mainstream technologies. Those are the technologies that not necessarily spell the leading edge, but they are actually the fundamental areas of our semiconductor industry. Those applications including automotive, it's including today's automotive safety, sensing, and the future's autonomous drive.
The IoT, as Bertrand mentioned, there's going to be 75 billion different devices connect to the internet. That's driving the fundamental of IoT from buzzwords three years ago to actually detailed applications today. Industrial automation and robotics. That's an area that people normally don't see the clear unit reflections, but it's actually driving a lot of semiconductors that are going into the industry. Security surveillance is another area. As we can see, we can easily list 10, 12, 15 different new applications that's driving our industry compared to the past. That's actually a big change of what's the fundamental for the industry. Again, if you look at this in the past 20-30 years timeframe, PC was always the bedrock of the industry. We call it the PC era, internet era, that's been driving the industry for a very long time.
Five, six years ago, smartphone replaced PC as the big driver. One driver become two drivers. PC is still there, smartphone kick in. 2017, I would call 2017 as an inflection year, because in 2017, suddenly all those new drivers, we're talking about 10, 12, 15 new drivers, kick into the industry, and suddenly drive the semiconductor explosive growth. We think all those new drivers are just at the starting point. If you look at each one of them, in the next 5-10 years, they're going to continue to grow. That's why fundamentally, we believe it's a great area for semiconductor industry, not because a technology is happening, but because all those drivers create the demand and need for a lot more semiconductors than we have in the past 10 years. What does that mean, fundamentally?
Fundamentally, just industry needs more wafers, because more semiconductor chips and turn out to be the industry has to make more wafers to meet those demands. This is a very simple curve of a million square inch, how many wafer starts does the industry have. As you can see, it's almost a straight line. You can exclude one of the financial crisis blip in 2009, and it's a pretty healthy growth trajectory. The reason that this curve keeps going up, and we believe it's going to going up faster and extending more in the future years, is that it's basically society is progressing. Society progressing, we need more electronics. In the past, we try to find which electronics can drive this curve. Is it PC? Is it smartphone? Today, you can list the 15 applications we just discussed. That's the fundamental driver of our business, more wafer starts.
We can look at some of the numbers of MSI. In the 10 years before 2017, 2006 to 2016, the CAGR is 2.7% for the MSI. The reason for that number, it's I think at the same level as GDP, is because lack of application drivers. As the PC decline, smartphone grow, and when the smartphone appear, PC already start to decline. There's a single driver that's supporting these numbers. 2017, 10% wafer start growth. A very fundamental shift. We look at it, what's happening, what exactly drive this high growth numbers. 2017 is actually not a good year for PC and smartphone. PC is declining 3% and the smartphone only grow 1.3%. The traditional driver for our industry has not contributing to this explosive growth.
This is why we believe all those new applications are actually become the factor for our industry, starting from 2017. Move forward, what's the future look like? 4.4%, that's what the third party forecast for the industry. You can see already it's almost twice as what happened from 2006 to 2016. As those applications continue to grow, we believe this number is very solid and could be even higher. Now, we understand there are a lot more wafers needed for the industry. What about per wafer base? We always talk about industry is progressing, there's Moore's Law. What does it translate to Entegris? Actually, if you look at two of the fundamental devices in the semiconductor industry, one is logic, the other one is memory. How much materials do they spend per wafer? On the left-hand side, that's a typical logic device.
A 28 nanometer versus the 7 nanometer that's going to ramp up this year. The material spending per wafer actually increased to 2.2x, 28 nanometer to 7 nanometer. On the right-hand side is the very exciting 3D NAND devices. Today, the 3D NAND production is at 64 layers, and the industry is migrating to 96 layer end of this year, and 128 layer in the next two to three years. Compared to 64 layer versus 128 layers, the material spending per wafer is going to increase by 1.8x. This is actually added the growth of demand for Entegris products, because not only the wafer start is growing, the material demand per wafer is also growing at a rapid speed.
If you look at some of the key areas Entegris is playing in, we acquired ATMI in 2014, and those are the key areas that ATMI products, or today we call it SCEM products, our specialty material products that play in deposition, which is the fastest-growing segment in the whole semiconductor industry, and this is also the fastest-growing business units for Entegris in 2017. We look at the growth for the next three years, it's growing almost 12% annually. For this area, that's because both the memory and the logic requires tons of deposition etch, deposition etch to achieve those devices. Copper and cobalt plating, those are the electroplating products that we dominant the market in the last 10 years. It's going to grow almost 8%, and it's also going through a transition from the copper to some of the new materials like cobalt. Specially wet chemicals.
We all know that every etch step is associated with the wet etch and clean step or cleaning step after that. As the etch step move forward, the demand for wet chemicals are also growing. It's growing about 8.2%. The dopant specialty gases, Entegris is famous for SDS, Safe Delivery Source, that's used for the dopant gases. It's going to grow 6%. If you look at all those things, Entegris is playing the top 3 of each of those markets, and those markets are all growing faster than MSI growth. That's because the material usage is actually adding to what the wafer start is growing. Now, there's a lot of questions about material usage, new material adoption is very exciting for the industry. What's the biggest challenge for our industry as we continue to progress on the technology?
Many customer will tell you it's the contamination control. Why is that? Because when you make a chip, those contaminants actually kill your chip, and it's affecting your yield. On the top is actually a spec that our foundry customers provided to us. They said from 28 nanometer logic device to 7 nanometer logic device, what are some of the key specs for contamination? The first one is the metal impurity. Metal impurity spec is actually reduced 1,000 times. It used to be parts per billion level at 28 nanometer. Today, at 7 nanometer, it's parts per trillion level. People even start to talk about PPQ, which is another 1,000 multiples in the future nodes. How to control those metal impurities in the materials, in the liquids, in the gas, it's becoming a very big challenge for the industry.
The second thing is the particles. We all know if a particle forms on patterns, it kills the circuit. Those particle spec has shrunk significantly. At 28 nanometers, the minimal particle size people control is 30 nanometers. Today, it's 15-10 nanometers. Extremely small particles to the point that many of those particles couldn't even detect by our current metrology. They challenge Entegris, since you are the contamination control supplier, those are the specs that we want you to achieve. What do those impurities or contaminants affect the wafer? We did a very simple calculation. For 1% yield, what does that mean to our customers? 1% yield, we always hear companies talking about 70%-80%. What about just 1% yield? What does that mean for one year?
The logic devices, for example, if you are making the key CPU or APU devices, 1% yield at a 10 nanometer node, is equivalent of about $150 million of revenue or direct profit for that fab. This is a 60,000 wafer fab, leading-edge logic fab. For the 3D NAND, making a 64-layer devices, it means $110 million. You can see the yield has direct impact on the bottom line of our customers. That is why they put so much more focus on controlling the contaminants, because contamination is the number one contribution to those yield impacts. If you look at Entegris is traditionally a number one solution provider for the industry for contamination control. We control the contaminants in the liquid, in the gas, and in the air.
This is our estimate of how each of those segments will grow in the next three years. Liquid, about 10% annual growth. Gas contamination control is about 6.3%, and air is about 6.2%. As the top provider of those solutions, we actually feel the demand from customers and also the challenge from customers to meet those industry demands. There are a lot of new applications coming into our industry. I just want to comment a couple of new applications that's directly being the focus today of the industry. The first one is 3D NAND. As we know, 3D NAND is driving a lot of CapEx spending, driving a lot of material spending for the industry. What our estimate is from 2016 to 2020, the capacity for 3D NAND worldwide is going to increase four times.
This is directly a result of the big data and also the need for better data storage solutions. As we all know, 3D NAND can hold large amounts of data at a very fast read and write speed. For Entegris, obviously, we provide the contamination control solutions, we support the fab construction and equipment, but we also develop a series of specific products for the 3D NAND. Some of those are the new deposition precursors that's used specifically for 3D NAND, very highly selective wet chemicals. As we know, 3D NAND, one of the key characteristics is super deep and very narrow. How do you clean those trenches and holes? It's become a big challenge, and Entegris is actively involved in developing some of those solutions through our partnership with customers. Then those devices obviously require particle and metal controls, very robust process tool components.
We also provide some of our product solutions, like coatings, to equipment makers to address this market. There is a big wave of updating their FOUPs. FOUP is actually the box to hold the wafers. Customers realize that in the past, memory always used less sophisticated FOUPs, but for 3D NAND, they need the best FOUPs available to the industry. A lot of opportunities for Entegris beyond our regular product pushes. The next one is EUV. There are a lot of discussions about EUV. EUV is basically a new lithography technology that enables the industry to continue to make the transistors smaller. After 20 years of a struggle, finally, EUV will get into production end of this year, next year. What does that EUV mean to Entegris?
First of all, we believe that EUV is very beneficial for the whole industry. It enables the shrink of the devices to the next generations. If without EUV, the industry, the Moore's Law will stop progressing. Overall, this technology enable the industry to continue to progress, and it's going to slow down the complexity increase, but it's not stopping the complexity. A lot of people say EUV is going to hurt material or equipment suppliers. We actually disagree. We think it's going to slow down the complexity growth, but the complexity is still going to be there. For Entegris, it's very unique because we are not only going to benefit the overall node shrink, we also have developed a series product addressed specifically for the EUV market.
Those products, including a reticle pod, it's a EUV reticle pod that's going to hold one of the most precious goods for the EUV technology, which is those EUV masks. Photoresist filters. EUV is likely to use a very different photoresist. In some cases, even including some of the metal elements. It requires some brand-new filters and purifiers for those resists, and our MC division's developing some of the leading-edge solutions and working with all the photoresist makers to implement those new filters and purifiers. EUV requires some of the post-etch solutions that our SCEM divisions are developing. Lastly, there are a lot of components in the EUV tools, and Entegris is providing many of those components to the equipment vendors.
One of the things that the whole industry is not paying enough attention that Entegris suddenly finally realized this is actually a very fast-growing segment, is actually we call the mainstream nodes. Those are the nodes at 32 nanometers and beyond. 65 nanometer, 90 nanometer, 0.18, 0.25. Those are the nodes that people don't talk a lot about. Those actually are the nodes that's driving a lot of growth in 2017, we believe they will continue to drive the industry in the next several years. The reason is, many of the key applications, the new drivers of our industry, like IoT, like industrial automation, like automotive application, does not require many leading-edge devices, but they require a lot of mainstream node devices.
That is why if you talk to some of the companies who have 200-millimeter capacity, they will say the utilization for 200-millimeter capacity is actually higher than the 300-millimeter capacity because those are the sweet nodes that providing the best cost for the devices that's needed for those applications. If you look at the whole industry, today, those mainstream nodes is about 49% of the overall capacity of the industry. It's going to grow, even though the percentage of it will decline a little bit, at 2020, it's still going to be 46% of the whole industry. It's in those nodes. As you know, those are the nodes that not buying equipment, not buying a lot of new stuff, but they continue to buy materials to make those wafers.
Entegris has relationship with many of those companies for 10, 15 years, suddenly we see a resurging of the revenue from those customers, we think this trend will continue for many years because of those application drivers. If we put all those three things together, more wafer starts, more materials, and contamination control, all three our divisions market, this is the same expansion of our three divisions. As you can see, 2017, we see the SAM is about $4.2 billion, it will expand it to about $4.9 billion in 2020. Expansion happens in all three divisions, for the materials, for the contamination control, and for the materials handling. As Bertrand mentioned, we look at overall 5%-8% growth. How does that put together? If we look at the overall society, the GDP is going to grow about 2%-3% annually.
The wafer starts, because of those new applications, the wafer start will grow faster than the GDP, adding another 1%-2%. You look at Entegris execution. What we see is there's material intensity favors Entegris. There's going to be contamination control that favors Entegris. Our execution and a product portfolio will drive the SAM expansion and also market share gain. You add all those three things together, it's a 5%-8% growth in the next three years. To summarize, market is great for Entegris. We will have more wafer starts, more materials per wafer, and contamination control will drive some of the key adoptions of our product. Put it together, it's a great setup for Entegris to continue to grow in the next several years. Thank you.
Thank you, Wenga. Really, I heard two things in that presentation. One, the semiconductor market's a better market than it's ever been before for three reasons. One, it's a more stable market. It's a more stable market because it's no longer dependent on one or two applications like PCs or smartphones, the breadth of the market. Two, growth in terms of unit production is accelerating. We're coming off a period 2006-2016, where wafer starts grew 2.7% into a phase where the period from 2016-2020, they're going to grow over 5%, so accelerating growth. It's a bifurcated market, meaning you've got the leading edge as well as the maintenance or trailing edge. If you're somebody like Entegris, who's been in business 50 years, that trailing edge is important.
We have products that are 20 years old that grew 10% last year because of what's going on, people sweating the assets in the trailing edge. That's the market overall, why the semi market is a really good place to be right now. When you think about Entegris' products specifically, he really talked about two things. He talked about the materials intensity and how when you move out an application, materials usage at some of the advanced apps is 2x. That's how you grow a specialty materials business 13% in 2017. He talked about particles still being the number one issue, in terms of defects in the fab. If you've got a contamination control business, that's perfect, right? That's how you grow a contamination control business 20%, you're tied right to that trend. With that, I'm going to turn it over to Todd.
Todd's our Chief Operating Officer, been with the company 22 years, had a wide variety of positions, everything from a product manager to running a sales organization, to running a division, and now the Chief Operating Officer. Todd?
Thanks, Greg. Good morning. Thanks, everybody, for coming to hear about our company today. It's exciting to talk about it, and you could hear it in Wenga's voice. He's pretty excited about the opportunities of the company, how the market is. I was really thrilled when he joined us, about six years ago. He came to the company thinking, "This is the company that's going to really benefit from what's happening in the industry in the next several years." He was really prescient in that case. He saw it coming and came and helped us actually plot a course to go through that. What I'm going to do is, move forward here.
When I started working for Bertrand about eight years ago, as he was COO, he gave us a pretty clear message that we're going to build a company that's going to listen to the customers of this industry and deliver what they need to succeed. Arguably, the most challenging manufacturing environment in the world is semiconductor, and to serve it is not a small task. You have to do a lot of different things. You've got to be relevant. You've got to be able to help them solve problems, and you've got to be able to support them as they go through these fast ramps.
I knew what I needed to do, as I became COO as well, is to make sure that we did the right things to build a company, to be a great partner and a trusted partner, to our customers, and to be resilient, to be able to, after 51 years, deliver a record year and plot beyond that to keep the growth going and keep the success going. I'm feeling very confident that we're ready to do that. I'll describe to you a little bit why as I go through my slides. These three phrases at the top here really reflect our customer's voice. This is what they say they need from their suppliers, and specifically from Entegris. Higher yields, as you just heard about, are money in the bank for them. They need to have higher yields quickly.
They need to ramp fast to high yields, and they need to be able to continually improve those yields as time goes on. Better performance really means we work at the very beginning of the process as they design semiconductors. More and more, Jim and his team are at the table helping them think through integration schemes to make those semiconductors become reality, reduce power consumption, increase speed, provide greater data storage and data access. He's helping to make all that happen, and that helps us get insight into what's going to come in terms of the fabs and what the equipment manufacturers are going to be working on as well to deliver that chip. The new one is reliability. You heard, both Bertrand and Wenga allude to things like self-driving cars, and as we were reminded even yesterday, reliability is going to be paramount.
If your PC goes down or your phone has an issue, that's an inconvenience for sure. If your self-driving car goes haywire, that's very bad. The latent defects is the new thing that people are starting to look for. When we talk about contamination control, it's not just do I get through the fab and I get out the door, it's what happens 10 years down the road when this thing's been heat cycled, if there's any latent metals in there that could cause a short or cross a boundary, they need to avoid those as well. When they talk about contamination control now, it's not just in the fab, but what's going to happen down the road. Things like purification to take metal ions out are becoming increasingly important. This is what customers have come to expect.
When I look at Entegris, I see two kind of main things that are really unique about us in helping to deliver that. The first is the diverse product portfolio. I'll talk about the three divisions. You've heard us talk about them before. They really help us go into a customer with much more relevancy to be able to help them solve problems in a lot of different ways. We've seen things in every part of the fab, in every part of the industry, including the supply chains. We know where contamination comes from. We know where issues come from. We know how to solve problems. We've done it before.
We can help a new customer, say a fab in China that's coming up, and we can help them get there because we've seen the problems that they're going to encounter, and they know that that's the value that Entegris brings. They invite us in early to be part of that discussion because of the breadth of what we can do and what we've seen around the fab. A diverse customer base. This is also something that's very unique about Entegris. You look at an equipment company, their customer base is really the device manufacturers. You look at a pure materials company, it's largely device manufacturers. If you look at Entegris, which is specialty materials and the contamination utilities around it, we serve the whole ecosystem. In fact, our customers, our large device manufacturers say we're kind of at the center of that ecosystem.
We can help connect everybody when we're working on joint developments, as Bertrand talked about earlier. It's a key to success. We have to be somebody they want to have there at the table early on to solve these problems, so we get the first shot at the business. Having that diverse customer base is a great education for us, and it helps us actually, "Hey, I know who to talk to at this equipment company. I know who to talk to at this chemical manufacturer. I know who to talk to at the wafer grower, and I can bring us all together and help solve those problems." Increasingly, our customers are asking us to do that.
That's built a company that helps us expand this served addressable market, as Wenga talked about, all the things that we do are finding increasing relevancy around different fabs and different devices than in the past, and that's helping us grow the company. There's some mission-critical things that are needed in the industry, and just want to highlight a couple of those for you. One of the critical needs is for being more than a supplier, being a problem solver really at any stage. I talked about that already, collaborating with customers from the device design all the way through to how they're going to implement that in the fab, and then how they're going to ramp and get to yield as quickly as they can. That's the second part of it.
This need to be able to go from an idea or a lab scale solution to high volume manufacturing on an ever-increasing slope, is really one of the natures of the industry now that we have to be responsible for. We have to place the right bets. We have to have the right capacities in place. You only do that right if you have a close collaboration with the customer, and you can kind of see what's coming, and that's really been critical for us. That ability to ramp up very, very quickly to high volume is increasingly challenging and increasingly important to our success. We built the company to really work on that. This is a little bit of a visual to help you see when I say kind of surrounding the fab, what do we mean by that?
If you look at the different divisions, as I'll describe to you a little bit more in a moment, we're really in every part of the supply chain and every part of the fab. From solutions for emerging technologies as they're designing these chips, as they're designing these advanced architectures that Wenga talked about and Jim O'Neill will share some more on, we're there helping them solve that problem, figure out how to do it. As they start to bring those materials into the fab, be they wafers, being brought in from wafer growers and our shippers, be it chemicals coming in and our chemical containers, be it our gases coming in and our dispense systems. Then in the sub-fab, as that gets distributed through the fab, refined, purified, filters, we're doing all that. All the way to the tool, to the point where it touches the wafer.
We talk about Clint Haris's business, the MC division, being the last line of defense before a chemical or material hits the wafer. The last thing it'll probably see is one of our products as it goes through the last purification and filtration stuff. That includes the environment around the fab, you see at the top there, including airborne molecular contamination, which is becoming increasingly important and very interesting problem to solve. If you go around the world to different fabs, everyone's in a different environment. The environment here in Massachusetts is very different from the environment in Portland than in Silicon Valley than in Hsinchu, or than in Shanghai. We have to go in there and find different things that are challenging that environment of the fab, and design a unique solution to it. We kind of see the whole picture around the fab.
When it comes to solving these yield problems, which are very complicated these days, may have multiple dimensions in terms of how to solve that. We've seen all of that, which makes us, again, a more valued partner, a trusted partner to our customers. Those are really important things about Entegris that are different than our competitors. That breadth, that ability to understand the problem holistically and to be able to actually come in with expertise and help them solve that problem. This is a little bit related, but I wanted to mention as well, because we do all these different things in the fab, we see different businesses of ours, as I'm going to talk about, ramp in different product lines, even ramp and then hit their long tail at different points in the fab life cycle.
I want to talk about that a little bit more when I talk about the divisions, if we're talking about our fluid handling products, as they're constructing the sub-fab and putting in the chemical environment, that's when we're selling products into that. When they bring in the tools and they're going to start to ramp the fab, that's when they start to bring in FOUPs. We'll start to see a big burst of business for those FOUPs as those come in. As they start up the fab, they'll clean it up, and they'll actually run a lot of filters to actually clean up the fab, clean up all the chemical lines, get ready for production, we'll get a big boost from there.
All those businesses then run into a long tail, where we'll see 5, 10, even 20 years or more of continuous business as they replace those products, as they consume them, or as they refit or upgrade the fab. All of them will have a long tail that actually lasts for decades. When I joined the company 22 years ago, we have product lines then that were leaders, that are still leaders today in things like fluid handling and filtration and many other parts of the company. When we grab that business, we have a very long tail that comes along with it. It happens at different phases in the fab for the different divisions. This is a little bit more data on how we're part of the ecosystem and we serve everybody.
The chart on the upper right is showing our sales by customer type. You can see, obviously, device manufacturer is very big. We're a consumables company. We have a lot of that as part of our portfolio. A lot of our sales do go to the fabs. They consume cleans chemistries as they put deposition materials down, as they filter and purify, as they bring in shippers with wafers and then chemicals. All that's creating business that we get at the fab. We also serve the equipment manufacturers. We make products that go into their tools, but also we help make sure those tools are designed to accept our products once they get into the fab, be it filters or something like that.
Wafer growers, as they're growing wafers, they use a lot of purification, a lot of chemistry goes into that, and of course, we make shippers for wafers. Chemical manufacturers, we are a chemical manufacturer, but most of the chemical manufacturers of the world are still customers of ours, and even to this day, work very closely with us on development for filtration, purification, and containers, especially. Those have become more important to them as well, because they see this contamination control becoming increasingly important. There's non-semi applications as well. We won't cover that as much today, but certainly what we do around contamination control, what we do in some of our specialty coatings and materials like graphites, are usable and are being used in other industries. We have some presence there as well. That's kind of why we're unique.
You're not going to see too many companies in this space that have a diverse customer set like that. Also makes us resilient. It makes us, not one part can go down if we see change in capital equipment happening or WFE. It has some impact, but not as much impact on Entegris as it might have on another company. Just expanding out the customer sales to the fabs to the bottom right there. This is an important thing to see as well, is that we've historically been largely logic and foundry has been the big drivers, because they've been driving the roadmap, until recently. Now, as you've started to hear, memory has become much more sophisticated, and they've started to need more sophisticated solutions as well. You can see our sales to memory there about 35% overall, 23% of that to NAND.
That's up from probably about 18% or so the previous year, and it's going to continue to grow probably 25% and maybe even more, going forward for the reasons that Wenga described. We've seen a nice growth in that. What we've done in logic has become very important to memory. That's really helped us to expand our SAM. These are the three divisions. I've described these to you in past sessions. Last year, we had each of the division leaders actually talk about them, but just want to mention kind of quickly, again, the nature of them. I won't read through the product lines, but each of them has several business units. The AMH, Advanced Materials Handling division, I'll start on the right here. That business gets its growth when there's fabs being built.
When construction is happening, that's when that business starts to sell fluid handling products into that. When that fab starts to equip and starts to ramp, they'll start to sell those microenvironments products, FOUPs, and we've seen a lot from both of those business units this year. That business grew very nicely, 10% for us, and a lot of it came from that stage happening in fab builds around the world. Liquid packaging for fluids and wafers, that comes in kind of on a continuous basis then as they ramp the fab, and they're selling at volume over time, we see that business grow then. Microcontamination Control, fastest growing business this year for us, for all the reasons that Wenga talked about, contamination control becoming more important to more types of devices.
They see a big push as the fab equips and starts to clean up the tool, and then they have a continuous stream of business as they replace filters at varying rates, depending on the application around the fab, for a very long time to follow. What's been happening in the builds of the fabs is very exciting for Clint and his business as they move into production and they ramp. That's when his business really starts to get going, both with the device manufacturers and with the bulk chemical suppliers that also have to use filtration. Specialty Chemicals, our largest division. This division is almost all unit driven, the exciting times are really ahead for it.
As these fabs have been being built are ramping, that's when we're going to be seeing more clean sold, we're going to be seeing more deposition material sold. Deposition material business unit is one of our fastest growing business units again this year, as I've said to you in past years as well. We're seeing a very nice win rate. We think we're gaining share in some of the specialty cleans as well, all of which will ramp when we see these advanced nodes ramp. We see good growth for that division as well going forward. A little bit more, before I jump into the divisions, I just wanted to mention, it's not just about growth.
You heard Bertrand say that one of the big focuses we've had has been on efficiency and financial performance, safety being first, not only for ourselves, but for our customers in terms of providing safe products to them. World-class customer service. We actually, even though we grew this year at a record pace, we focused on service level and trying to make sure we're improving our on-time delivery in spite of all the pressure on us to deliver more and more. We actually achieved that in 2017, improved our service level. Quality and stability, we've talked about in the past. We've reached a 5 sigma level of quality, which has been a long journey for us, and we have to continue that going forward. Efficient manufacturing. When we talk about margin improvement, it's really about that.
It's about taking the infrastructure that we've had, continue to grow, and fill it out as we've been doing this year. That'll continue to give us some volume leverage. We're also going to see some mix of benefit. If you look at the growth I'm going to show you, the unit-driven businesses are growing the fastest. MC and SCEM are our fastest-growing divisions. As you can see, they're the higher margin businesses as well, that's going to help us from a mix perspective. We have a lot of continuous improvement initiatives going on. We've been able to leverage OpEx appropriately as we've gone through this growth, very quick growth lately. All that's designed to deliver, as Greg will describe, improved margin performance by 2020 of about 300 to 400 basis points. Very quickly on the divisions, I've already talked about them a fair amount, especially chemicals.
2017 grew 13%. Again, that was against an MSI growth of about 10%, outperformed by about 300 basis points. Operating margin reached 27%. We expect this business to continue to grow at 200 to 300 basis points above its market. MSI of 4.4, it'll outgrow that, 6% to 7% overall growth in the margins. This is about a 200 basis point improvement in our outlook for the margins for this business as compared to what we showed you last year. We're increasingly optimistic about the business. We've talked about the growth drivers and the margin drivers, as I talked on the previous slide. Microcontamination Control. Sales growth 20%, a very strong year for this business. All the drivers that Wenga talked about, but we also unlocked some new capacity.
As you hear us talk in the past about our i2M center down the road here, which is the membrane manufacturing that's fully online, we were able to ramp that completely and get that and other investments, as you'll see if you go on the tour today in our manufacturing capabilities here. That helped unlock some of the growth in MC on top of the market. Again, a 10% market, 20% growth. A lot of that was being able to just really catch up and deliver more and more than we ever have. Margins very strong at 37%, you can see our outlook there to continue to grow 200 to 400 basis points above the market. We've actually taken this adjusted operating margin range up about 500 basis points from last year. We are getting the leverage. We're filling out the factories.
We spent the time or spent the money to invest in this business several years ago for R&D and manufacturing, it's starting to pay off as we fill that out in better financial performance. Advanced Materials Handling. I've got a long history with this business. When I joined the company, I was actually part of what's now this division. Grew 10%. We were happy with that this year. Again, it's really about where they're at in the fab build cycle. We saw very strong demand for the fluid handling products and the advanced FOUPs. We actually are adding capacity in the FOUP business right now to keep up with continued expected strong business in that throughout 2018 and beyond. We don't expect it to grow as much. It's really more of a flat to 100 basis points above market.
We've really focused this business on margins. You remember us talking about this last year. We went to work really mid-year this year on improving margins for this business, made a lot of choices about what we're going to invest in for R&D, reducing our footprint a little bit for this business, improving the factory utilization. We're still dedicated to hitting our margin range of 22%-24% by mid-year, and I believe we're on track to do that. Our second half margins for this business were stronger than the first half last year, so we're on the right trajectory to make that happen. Can't not talk about China, I want one slide on China here. We had very strong growth again, 26% this year in China.
It's really coming from, right now, this new fab construction that's been happening the last couple of years. We've done very well with our fluid handling products, wafer handling products, filtration, really across the board in China, because we're that unique partner to those kind of companies. What's ahead of us is the good news is those will start to ramp, and we'll start to see high volume manufacturing. Consumables will start to be consumed at a greater pace. Our strategy really hasn't changed. We've been investing in local talent, obviously, and the infrastructure to support the market. We've had these sourcing partnerships, which we've talked about, and we had a couple of them in the last 12 months around especially our SCEM division. Just market coverage, having enough salespeople in the right places.
The new thing that we'll talk about a little bit more is the local applications lab. As we have in Taiwan and Korea, U.S., and several locations, we have applications labs to help our customers pick the right product quickly and get the product installed and working effectively. We're going to construct one of those in China. We started the process already this year. I'll conclude. I think I've hopefully described to you what we've constructed as a company here to be responsive to these needs that Bertrand and Wenga outlined. It's really about having the capabilities to work across the whole ecosystem, becoming that trusted partner where we can come in there and work from them from the very beginning of the chip design through long-term running of the fab and continually looking for improvements in operating efficiencies in the fab.
All that's produced these long-tail revenue streams that keep the company very resilient to changes in the industry. Right now, a lot of things are coming our way in terms of the applications and the demands on the company, we've really, I think, set it up well to succeed going forward. Thank you.
All right. Thank you, Todd. When I listen to Todd, I think about two things. I think about diversity, diversity and capability, and capability. Diversity across the supply chain, diversity across the customer base, diversity across product lines. That really distinguishes us from many of the other companies in the space. We're not tied to a single group of customers. We're not tied to a single group of product lines. It's like your investment portfolios. Broader is better from a risk and mitigation standpoint. I view that diversification as being a risk mitigator. Todd talked a little bit about capabilities. In today's market, the ability to have advanced quality systems, the ability to ramp a product line quickly, the ability to have the channel to market that we have are all things that distinguish us, particularly as we move to the advanced nodes.
Smaller competitors are essentially out of the game because they don't offer those capabilities. That diversity and those capabilities are really what make us a trusted partner. The diversity gives us great applications expertise within semiconductor. The capabilities are obvious to the customer. The other thing I would point out is the capabilities are also what allow us to expand the margin. Getting better and better at the things we do is what allows each of those divisions to expand their operating margins. With that, before we go to the next presenter, we're going to take about a 10-minute break. The lunches are just outside the curtain. Try and be back in 10 minutes or so, thank you.
I just want to give everyone a two-minute warning. We're going to start in a couple of minutes with the next presentation, if you could make your way back to the room and your chairs, we can start immediately. Thanks. Okay, everyone. I think we're going to get started. Greg, do you want to introduce Jim?
Jim O'Neill is our Chief Technology Officer, been with Entegris about five years. Before Entegris, 23 years with IBM. Those of you who have been around the industry know that IBM developed the process technology, which is the backbone for much of the industry. He oversees our CTO organization, but also sort of guides our entire portfolio management process. Jim.
Thanks, Greg, and good afternoon, everybody. I want to reiterate point that Todd made in his presentation, in that one of the key differentiators for Entegris is the breadth of our capabilities. Whether it's performance materials or yield-enabling technologies or purity-enabling materials handling solutions, Entegris really occupies a unique niche in the industry relative to our competitors as a provider of what I would call holistic solutions that really go after some of the most challenging yield and performance problems that our customers face in the industry. What I hope to convey in the next several minutes here is a story about how Entegris thinks about innovation from the context of how we leverage this breadth and pull our pieces of technology together into more holistic solutions.
As you're aware, for more than 50 years now, Moore's Law has guided the industry as device manufacturers have worked to improve technology, increase density, increase performance. These advancements historically have been achieved through a process of miniaturization through advancements in optical lithography. If you think about today, the fact that structures on advanced logic devices have dimensions that are less than one tenth of the wavelength of the light that's used to create them, this is a pretty remarkable accomplishment. Such lithography advancements come with added complexity and increasing difficulty and increasing costs, leading some to conclude that Moore's Law can't continue, or at least it will slow down.
However, there are a number of clever innovators in the materials front who have really figured out ways to leverage the base performance of materials and use these materials in advanced device builds to continue to improve performance, and allow the industry to keep pace with Moore's Law. Really, at this point in time, materials account for the vast majority of the performance improvement that we see in advanced node technology from one node to the next. Really, this is a story that's all about materials. The reliance on improved materials to keep pace with Moore's Law has really introduced a new set of complex challenges. More materials means more process steps. More process steps means more complexity and more challenging yield ramps.
For example, if you consider the number of process steps in a seven nanometer foundry logic technology, it's approximately twice that was found in a 20 nanometer logic technology several years before. The same time, the technology introduction ramp at many of the leading-edge fabs is accelerating dramatically. For example, the production ramp for a 10 nanometer logic part is about six months faster than the ramp rate for the previous generation 16 nanometer technology in the same fab. We've got a more complex technology ramping in a faster period of time. We've got a real challenge in the industry. Our customers are focusing and pushing us to help them with speed to yield, and this is an area where Entegris plays very strongly. That's really the environment in which we operate.
In this challenging environment, our customers are looking for material suppliers that can provide an increasing array of capabilities to support them. In particular, they're looking for three things. The first is a comprehensive set of engineering capabilities to solve difficult problems. Secondly, they're looking for the technical expertise and skills that they can engage with on par with their engineers and scientists to collaborate in the solving of these problems. They're looking for material suppliers that can innovate and execute quickly, given the challenge of the node ramps that they're facing. I would argue that Entegris is responding to all of these. We are increasing our investment in three main areas. The first is in processes that help us accelerate our new product development.
The second is in people and skills, which allow us to engage directly with our customer in the solving of these difficult problems. The third is in the infrastructure that supports our customers in the geographies where they operate. What we're actually doing here is building out our global technology footprint. I'll talk about each of these in series. The first is the process. If you consider Entegris, we have a portfolio of more than 20,000 products, and we have a history of more than 50 years of innovating solutions to difficult yield challenges for our customers. We have a long history and a robust culture of innovation, but we've continued to focus on areas where we can continue to improve. Recently, we've implemented the new unified stage gate process for our new product development.
This was implemented after the acquisition of ATMI, and is really focused on providing the right data at the right time to enable us to make timely database risk-adjusted decisions on how to proceed with a project or portfolio in our development cycle. Secondly, we've greatly increased our focus on our comprehensive portfolio management process. This is a process that is formally executed quarterly and looks at all development activities across the whole corporation. What this allows us to do is to prioritize our activities, rebalance resources, and really get good headlights into what projects are coming and what products are coming down the funnel, and when. Two years ago, we implemented a corporate-wide series of innovation summits. We've held these in the U.S. as well as in Asia.
The focus of these summits is really to bring technologists from across the company together and identify ways in which we can truly leverage the breadth of our capabilities, not only across divisions but across geographies. The output has been a series of cross-divisional, technical, strategic focus areas, one on 3D NAND, which we've talked a lot about. This is actually expanding into solutions overall for advanced memory beyond 3D NAND. Secondly, we've focused on EUV lithography or extreme ultraviolet lithography, and this has expanded into a program on advanced solutions for patterning and overall advanced lithography. Also we have an effort on wafer level packaging as wafer level packaging simply extends the process environment from what's traditionally the back end of the line.
What's important about these cross-divisional initiatives is that it forces us to really look at things from the perspective of the customer, who frankly doesn't really care which division a solution comes from. It's important that through these, we build the concept of a one Entegris, and look at the solution set that we're providing for the overall customer's problem. Secondly, we're investing in our people and skills. Entegris has inherently a highly skilled technical workforce. About 35% of our R&D team has a Ph.D. level degree with an equivalent number of master's-trained scientists and engineers. Increasingly, this workforce is positioned to face the customer through collaborative projects and joint work in our Entegris technology centers around the world.
We've put particular focus on strengthening our Asia workforce over the last several years, as this is where a number of our technology centers and a large piece of our customer base is located. The net result of all of this is that a given account manager really is positioned and can in fact draw on the whole strength of Entegris, if need be, to solve a difficult customer challenge. Finally, we're continuing to invest in our infrastructure. Last year, Entegris spent over $10 million on improving tech centers and improved equipment. Some examples of some of the activities that are either recently completed or currently underway, as Todd mentioned, we are installing a lab in Shanghai for applications developments within our China market.
We are in the process of installing new metrology equipment in Taiwan, which will give us improved capability for on-wafer particle detection, at a level which is on par with what our customers have in their most advanced fabs. Our ability to generate customer relevant data on the performance of our materials and chemistries is greatly improved. We've implemented a new chemical mechanical planarization lab in Taiwan. This is a CMP facility to develop both clean formulations and do application work with our customers in Taiwan, but also more broadly in the Asia landscape. We've built two new filtration labs, one in Taiwan and one in Korea, again, for both advanced filter R&D as well as application development. Something you'll see a little bit later today is we have built out a new analytical lab here in Billerica to strengthen our overall characterization capabilities for contamination and defects.
The focus of all of this investment is to build Entegris capabilities in a manner which best serves our customers. With this skilled technical workforce and these enhanced engineering capabilities, Entegris is really positioned to work both collaboratively and on par with the engineers and scientists within our customer base. This is really important if we're going to really, truly understand the nature of the challenges that they face and develop more complete or holistic solutions to those problems. We work very closely both with the equipment supplier community and the end user device manufacturers to implement our solutions. It doesn't really matter whether it's cleans or deposition precursors or FOUPs or whatever. We're the only company that can really bring such a broad range of not just products, but solutions to the table, to work on these challenging problems.
Let me give you a couple of examples of some work that we've done in the area of 3D NAND. On one project, we have been working for some time with a memory device manufacturer to provide a new deposition precursor to deposit a very critical film at the heart of the device in a way that will both improve performance and improve yield. We have identified and brought to the table a new series of precursors which will satisfy this purpose, but we've also brought a suite of engineering capabilities which really serve as the infrastructure to surround this chemistry or this molecule and allow its defect-free delivery directly to the wafer, so we can provide the molecule, the delivery vessel, the special coatings on the vessel to ensure pure delivery, as well as a low vapor pressure filter to enable to further reduce contamination.
We have, in fact, been able to provide not just a material, but a holistic solution to this deposition problem. That's an example of what I mean by pulling the broad elements of Entegris together to provide a more complete solution than that our competitors can do. A second example with a different memory device manufacturer and an equipment supplier, in this case, is intended to provide an etch formulation for the selective removal of a particular film in a very complex stack of materials, again, in the heart of a 3D NAND device. Entegris, our Wet Clean R&D team, developed this formulation that performs according to the customer specs.
Rather than just providing the chemistry, we've also provided the appropriate filtration and the container solution that allows the installation and delivery of this clean chemistry cleanly to the wafer so that it can be fully utilized. Again, we're able to provide a holistic solution for the customer. In doing this, I think we've established ourselves as more than just a materials provider, but really a true partner in the innovation process. This innovation engine is really beginning to bear some results. Entegris is increasingly an innovation partner, we've increased our investment in base research, and so over the last 18 months, we've seen about a 10x increase in the number of fundamental R&D projects that we've kicked off.
These projects are really a reflection of our recognition that we need to continue to increase our base understanding of fundamental interface problems and chemical problems that are really at the heart of what our customers are challenged with. Secondly, because memory now encompasses an increasing portion of leading-edge challenges, we've increased our R&D focus on memory projects. Over the last three years, we've seen a 6x increase in the number of memory-focused R&D projects. Historically, we've focused on logic because that's where the leading-edge challenges were. Now, with an increased focus on memory, especially 3D NAND, our portfolio is much better balanced. We're having some success with these new products. This is leading to growth in our new product revenue. New products now comprise about 35%-40% of our overall revenue. This is a five-year NPR basis.
We think that at this level, we can sustain a level of innovation within our new product portfolio that will be able to support our growth objectives. We think the innovation level is about right, and it's being reflected immeasurably in our revenue results. I hope what you gather from this is that Entegris' success is really dependent on innovation. We occupy a unique niche as a company with the breadth to provide broad solutions, complete solutions. The challenge on us is to innovate in ways that really leverage the breadth of our portfolio and provide these combined, these comprehensive solutions that really satisfy our customers' needs. For that reason, we're primarily focused on materials innovation, holistic solutions, and strategic partnerships or collaborations with our customer base to enable Entegris to best serve our customers and continue to grow. Thank you.
All right. Thank you very much, Jim. Take away from what Jim had to say to us. First of all, we're absolutely committed to innovation. I mean, it's the lifeblood of this business. Our customers want to innovate. We want to innovate, help them innovate. Secondly, it's really an extension of what Todd talked about. It's all about capability. You'll notice that Jim didn't spend any time talking about this product or that product, but he talked about how we do things and our capabilities, because our capabilities are going to be what allow us to sort of regenerate our pipeline. He talked about those in three different ways. He talked about people, 400 or so R&D people, a third of them with PhDs, two-thirds of them with advanced degrees.
He talked about the investments that we've made in our tech centers and our applications labs that are close to our customers, that allow us to help customers solve problems. The last thing he talked about, what I want to hit on a little bit more, is process. What we've done from an R&D process perspective, and really, when I say we, I mean Jim and his team, is we have a new product development process that spans the organization. The technology people don't all report to Jim, but they're all expected to use his tool. That tool has helped us prioritize. Every new product has stage gates. It's even got for a finance guy an ROI, which is like, that's great. The point is, we've become much better.
You look at the money we're spending on ER&D and the nature of our pipeline, we've become better at what I call R&D investment. We've made better choices. We've put more money on the bigger opportunities, and it's paid off when you see the growth in what we're doing in 3D NAND and the fact that our R&D portfolio has six times as many projects tied to 3D NAND as it had a few years ago. Those capabilities are really sort of what's going to drive us forward. We're going to switch over. I'll talk a look to the finance part of the discussion. Before I get into my presentation, I just want to do two things. One, I want to thank my colleagues. Bertrand and I, it's in our job description to do these things.
The other three guys have day jobs, and it's a big effort to do this, I want to tell them I appreciate it. I also want to tell our corporate marketing team who helped us this year. They came in and they looked at our decks from last year, and they said, "Wow, you guys can do better than this." I think we have done better with their help. Obviously, Steve and Jolene and Gina on the IR team have been a big help here. With regard to finance, I'm going to talk about three things. I'm going to level set a little bit on the objectives. I'll give you a little bit of historical perspective. I know I only get a little bit on that because that doesn't matter. It's like, what are you going to do tomorrow?
I think you've heard a lot about that. We'll provide an update on the target model, then we'll provide an illustration around what that target model can mean and what that can mean with a couple of different capital allocation overlays. First of all, from an objective standpoint, I think it's pretty clear. We want to continue to outgrow the market, the market being defined as millions of square inches of silicon produced by 200 to 300 basis points. Operating margin expansion, Todd laid it out. Each of the divisions has expansion in their margins. Corporate-wide, our goal over the three-year time horizon is a 300 to 400 basis point improvement in the EBITDA margin. Continue to improve our free cash flow yield. That's really important because cash is ultimately what we have to allocate toward investments and what we get paid to do.
Then finally, we want to continue to drive higher earnings per share. Lower right-hand side of the slide, strict financial discipline to achieve the target model. I kind of hate the word strict. It reminds me of the third-grade parochial teacher. That's not what it means. In this vein, it means primarily that we're disciplined around what we do from a financial perspective. That doesn't mean that we have a lot of tension in the organization and people are saying we're under-investing. It means we're thoughtful around where we invest. We spend a lot of time deciding how to make choices. We spend a lot of time on capital allocation. Talking about the results. 2017, year of many records. Revenue up last year 14.2% on a CAGR basis since 2013, 18% growth in revenue.
Put that in perspective, we always talk about, you could say, well, a rising tide floats all boats. We talk about outperforming the market. Since 2015, which was the first full year of operation since we had ATMI, our market has grown 8.8% on a CAGR basis. We've grown 11.4%. We've outgrown the market by 260 basis points. The 200 to 300 that we talk about going forward, not an aspiration, something that we've done before. EBITDA and EBITDA margin. EBITDA margin last year, it says it exceeded 25%. Fact is it was in excess of 25%, it was almost 27%. It's grown 28% on a CAGR basis since 2013, and last year was up 35%. Revenue up 14%, EBITDA up 35%. Is there leverage in the model? I think so. Earnings per share up 53% year-over-year.
We've talked about making a buck forever. Last year, we broke through that ceiling in a big way, $1.44. Again, leverage all the way down the P&L, 14 on the top line, 35 on the EBITDA line, 53 on the EPS line. Very good growth, 26% CAGR since 2013. Free cash flow, same story. Free cash flow margin up about 280 basis points from 12.1% to 14.9% last year. On an absolute basis, free cash flow was up approximately 40%. We think we can continue to expand that margin. Something to point out with regard to last year's results, say, well, the earnings were up 53%, the EBITDA was up 35%, you'd expect free cash flow to grow 40%. Our CapEx was up last year by almost 50%.
In the face of a 30% increase in our capabilities and our capacity, we managed the balance sheet well on the working capital front. DSOs improved. Inventory turns improved. It wasn't just, oh, better earnings, better cash flow. We invested more in the business, we were careful as we managed, like I said, the working capital side of the business. Shifting to the balance sheet. We exited the year with a net leverage ratio of 0.1 times. 2014, when we acquired ATMI, we took gross leverage up over 3, net leverage was at about 2 times. The balance sheet at year-end, $625 million in cash, about $275 million of that was in the U.S.
Our long-term debt, $680 million, consisting of the $550 million of notes that we issued in the fall, and about a little more than $130 million of the term loan that remains from the ATMI acquisition. We've repaid, since the acquisition, $326 million and expect to pay another $100 million over the next 12 months. When you think about the balance sheet going forward, we'll continue to focus on liquidity and being in a position to be able to do acquisitions. Also think about that $550 million as our permanent level of debt. That represents call it 1 to 1.5 times EBITDA, 1 to 1.5 times turns of EBITDA is our permanent debt. Bottom line here, though, is very sound balance sheet, fair amount of flexibility as we move forward with our strategy. Capital allocation.
Bertrand talked about that, and I just wanted to provide some historical perspective. We do talk a lot about acquisitions, and we think that is our best opportunity in terms of capital allocation. Think about that in perspective of what we've done in the last five years. We've spent $843 million on acquisitions. The vast majority of that, a little over $800 million, related to ATMI. If you think about what we've done from operating earnings, EPS, cash flow since that ATMI acquisition, hard to dispute that investing in acquisitions isn't a good idea. We've invested $463 million in ER&D over the past five years. That number between 2013 and 2016 as a percentage of sales moved up pretty consistently. 2017 was flat with 2016, yet we continued to improve our new product pipeline. We continued to generate good returns on our new products.
Moving forward, I would expect you to see, in absolute dollars, an increase in what we spend on ER&D. Debt repayment, we talked about that already, $326 million since the acquisition of ATMI in 2014. CapEx, $349 million over the last five years, $90-plus million last year, $60-plus million the year before. We'll increase that number again in 2018 to somewhere between $100 million-$110 million. So where have we spent that money? We've spent that money on membrane capability. Wow, did that show up in the P&L when you look at what we did in Microcontamination Control. We've spent that money outfitting a facility in Korea. Look what we've done in Korea the last few years. I think we've made good investment choices on the capital side, and we'll continue to do so in the coming years.
From a shareholder return perspective, about $10 million a quarter going back to shareholders in the form of a buyback, $10 million going back to the shareholders a quarter in terms of the dividend. $80 million a year going to shareholders. When we look at it over the last five years, the numbers aren't so big, because like I said, we just implemented the $10 million a quarter in Q2 on the buyback, and the dividend was something new for us in the fourth quarter of last year. When you look at the chart historically, not so big, but if we look at that chart next year, you're going to see a different picture in terms of what we've returned to shareholders. I want to talk about two things on this slide.
One is I want to give you my perspective on the portfolio, and two, want to talk a little bit about we're going through an accounting change in the first quarter of this year, and I want to highlight that so nobody's surprised when we come out of Q1. First of all, we are very fortunate in terms of the portfolio and how it ties to industry trends. All four of the presenters talked about the increases in materials intensity. Specialty Chemicals and Engineered Materials is tied right to that trend. Wenge talked about the impact of particles in the fab environment and contamination control in the fab environment. Microcontamination Control, right on trend there. AMH, largely around contamination control.
Each of the divisions has an expanding margin profile. We're not reliant when we talk about 300-400 basis points of margin improvement, it's really across the portfolio. Biggest improvement in Advanced Materials Handling, where we were at 16% last year and moving up 200-400 basis points over the horizon. Let me talk about the accounting change. The center of this slide really tells the story. You've got the adjusted operating margin. You've got an old number and a new number and a 3-year outlook. The old number, if you go and pull our financial statements from last year, that is the operating margin of those respective divisions. In 2018, historically, we have been very pure in terms of our centralized cost.
We've said it's not fair to charge a division for IT because they don't have a lot of choice around what IT system they use. We choose that centrally. It's not fair to charge them for accounting. We decided it really is fair to charge them for all of that because all of our peers do. We've got a bucket of unallocated corporate costs when you look at our segment data, about three-quarters of which we're allocating back to the divisions moving forward. What does that mean? It means for each of the divisions, their published results on average will be about 400 basis points lower than they were historically. It means nothing on a consolidated basis. The 3-year outlook is relative to that new number.
If you try and tie my 3-year outlook to Todd's 3-year outlook, it's slightly different, but this is what you can expect. There's no change in the relative nature. Each of the divisions still has an improving margin, greatest of which is in AMH. Long-winded, but I just want to make sure everybody understands that when we come out of Q1 and we say, "Hey, our MC division had a 33% operating margin," people won't say, "Well, wait a minute, it was almost 38% last quarter." It's not apples and apples. Let's talk a little bit about where we are from a target model perspective. This page really breaks into three sets of numbers. On the left-hand side is sort of historical perspective, 2016 and 2017 performance.
The middle of the page is our current external model in terms of where we expect to operate. The right-hand side is our expanded external model, which takes into account higher revenue levels as well as some of the operating leverage. Let's just focus on the right-hand side. At $1.5 billion in revenue, we expect our adjusted operating margin to be 23%, our adjusted EBITDA margin to be 28%. When you look at analyst estimates today, most of them are in the $1.475 billion range. That's the kind of numbers we should be looking at for this year. As we move out and apply the growth that we've talked about, we do get to that 30% lower right-hand side, 30% EBITDA margin and $2 per share plus in EPS. Let's take everything you've heard today and sort of put it into an illustrative model.
This chart, I want you to focus on two things. One is the upper left-hand corner. This is organic. This is what we can do without M&A, without buybacks, just driving the business. In the lower right-hand corner, where do we get to in terms of an operating margin or an EBITDA margin at 30% and $2 plus. Revenue, $1.343 billion last year. At a 7% CAGR over the next three years, that number becomes $1.646 billion. That 7% ties with everything you've heard today. Down at the bottom of the page, 3% GDP growth, call it a point and a half of growth for semis ahead of GDP, that's 4.5%. When you talked about 4.4%, we outgrow that market by 2%-3%. 7% top-line growth, 300 to 400 basis points of margin expansion.
The numbers, like I said, it builds on everything that you've heard today. The key here, that's organic. That gets us to $2-$2.10 a share, and everybody's saying, "Bertrand said $2.50. Where is that?" Let's go and talk a little bit about what does capital allocation do for us. Four columns here, four different capital allocation scenarios. On the left-hand side of the page is retain cash build liquidity. That would have no impact on our earnings per share if we just built the cash up. Second scenario is what if we just took everything in excess of $150 million in cash and repurchased shares? That would provide accretion of $0.35 a share. Put that on top of that $2 or $2.10 a share, and you get very close to the $2.50 that Bertrand talked about. Another alternative, obviously, is M&A.
M&A, again, take all of that free cash flow, all the cash on the balance sheet above $150 million, buy companies at 12 times EBITDA. We would get $0.50 of accretion over the next three years. Now you're talking about a scenario where you're kind of $2.50, $2.60. The last scenario says, yeah, but, we've got this great balance sheet, and we could potentially lever that balance sheet. That scenario is what if we found a number of decent sizable deals, which we think we will do, and we actually use some leverage for M&A? Reasonable leverage like we did in ATMI, three and a half times. That scenario would provide $1 of potential accretion. That would take you to a number, call it $3-$3.10.
The last two slides roll everything together that we've talked about. In summary, I think we're well-positioned from an organic growth perspective, no question. We think with thoughtful capital allocation, the number we threw out as a commitment is $2.50. I showed you a scenario where we could potentially do better than that. In summary, at one level, very simple story. We're operating in a market that's as good as it's ever been, and there's lots of reasons to think that it'll continue to be a great market. The products we sell are right on trend. Increasing in materials intensity, increased contamination control requirements. We've got a really good balance sheet to start out as we move forward, and we've got a history of really good execution. For all those reasons, as a team, we feel great about where we're going at Entegris.
With that, I'm going to bring the rest of the group up to answer any questions.
As everyone is coming up on the stage, I just want to remind the people in the room that if you are interested in taking our tour, we're planning to start that at around 1:30, and we'll meet in the back, near the windows. Switching the slide up.
Thanks. I think my question, Edwin Mok with Needham. I have three questions for you guys. First, on DRAM, as you've shown on your slide, that's the smallest part of the three end market that you guys have highlight. But at the same time, we hear more logic processes going to DRAM, I think there's a lot more tungsten and depth process going to DRAM. Given that you've put all the work into NAND, have you started to see some momentum gaining, and you expect that to grow at a faster pace? I have two more follow-ups, sorry.
I would take the first part of that question, and then we'll turn to you, Todd, or Jim, or Wenge, if you want to add. What we are actually saying is, first of all, is we are allocating a lot more R&D spending to the memory technology roadmap. That includes actually advanced DRAM as much as it includes 3D NAND. It's true that traditionally, the DRAM architecture has been a little bit more forgiving and provided less opportunities for the types of solution that we're providing. I think this is changing, and this is changing actually really quickly as the DRAM transitions to the 1X node. I would expect actually those transitions to be very, very beneficial to Entegris, and that should actually translate into our revenue for DRAM makers going forward. Todd, Jim, Wenge, do you want to add anything?
I think that said it actually completely. The only thing I'd add to it maybe is that as the memory manufacturers started to work with 3D, and they started to have those challenges, they learned more about contamination control and about things like advanced FOUP wafer handling with advanced FOUP technologies. They're starting to look now to bring that back into the more advanced DRAM as well. I think that exposure we got is actually going to help us proliferate into DRAM. We feel quite optimistic that DRAMs come alive in terms of needing the kind of things that we do.
I think there's one aspect that's interesting, which is stacked memory. I think it sort of leverages both memory trends, also wafer level packaging kinds of applications as well.
I would just adding from the market point of view, 3D NAND has added a lot of capacity in the last year. As you can see, our revenue's not only from the materials in the fab, but also from the tools and wafers and others. That's why the 3D NAND revenue is growing very fast. DRAM has not been adding capacities until now. As they start to invest more on the capacity, I think we will see the benefit not only from the direct revenue from the fab, but also from the equipment, from the materials and others.
Great. I have two questions for you, Greg. First, if I look at the difference between your target EBITDAR and operating margins, they will stay at five points even though you raise your revenue by $300 million. Does that imply you're going to accelerate your CapEx? I just want to make sure.
Yeah, if you look at our depreciation today, it's running in the $70s, and we've been investing. Last year, we invested in the $90s. This year, it's over $100. Over time, yeah, that depreciation is naturally going to increase.
We expect this.
Yeah, to the question, do I expect that at $100-$110, I'm not expecting us to accelerate from there next year.
Thanks for clarifying that. Then on your capital allocation chart, you show that on a leverage M&A, you can potentially get to a $1 incremental earnings from that, right? That seems like a really big number. If I did my math correctly, that's like a over $1 billion revenue company that generating. Like you said, if you're shooting for 12x EBITDAR on an M&A, right?
Well, it's a combination, right? If you think about where our leverage sits today, it's a very significant amount of debt that we could borrow. Our annual cash flow, the profile of that cash flow is different today than it was 12 months ago, because 12 months ago, I'd have told you about 20% of our cash flow is good for M&A. Today, I would tell you the vast majority of it is because of the change in the tax law. The capital that we have available to allocate, the picture's quite different today than it was 12 months ago.
I think what we're trying to do with this model, Edwin, is really to actually ask Wall Street to start really incorporating some element of the capital allocation strategy to the valuation of Entegris. I think that the current valuation of Entegris primarily reflects what we'll be able to do organically. I don't think it necessarily reflects some of the flexibility that we have with the balance sheet. I think that we have demonstrated as a company that we are very thoughtful allocators of capital, and you should expect us to make the right decisions, whether that takes the simplest form, which is a buyback or something actually more significant, as Greg was describing. We're not pointing to any one of the options.
We're just saying that if you look at the panel of options that we have, we should be able, at a minimum, to create an additional $0.35 of EPS power, and maybe on the outside of that, up to about $1. That's really the takeaway.
It's Sunil with Deutsche Bank. Thanks for doing the presentation. I guess I'll start off with a question for the team. In terms of operating margin, is there a path to 30%, which some of your competitors are doing, or is it just not realistic based on your revenue growth targets?
Well, yes, there's a path, and I think that if you look back at the walk forward that Greg was presenting, it clearly says that we should be in a position to reach that 30% bottom line performance, EBITDA performance, in three years from now, assuming
I'm talking about operating margin.
Operating margin. Okay.
Well, I would say we have peers that are 30% EBITDA margin. We don't consider Lam Research a peer. Our peers are Cabot, Versum, and those guys are, in one case, over 30%, in one case, approaching 30%.
Okay. Maybe staying on the operating margin side, maybe by segment, if I add back that 400 basis points difference, and just for comparison reasons, it looks like the contamination control side is going to have the biggest improvement from last year's, maybe by 500 basis points, and followed by maybe 200 basis points from the specialty chemical side. Just trying to understand what is driving that kind of improvement. I have one more.
Yeah. The biggest improvement in margin this year is actually going to be in the AMH business, because that's the business where we've right-sized the cost structure, we've exited some businesses, we're closing a facility. There is leverage in the MC business, it was 500 basis points last year, but it's not 500 basis points in 2018.
Yeah, I'm actually asking about the target operating margins a year ago when you, at the Analyst Day, have a certain set of numbers. They are not comparative to this year's because of changes.
Right.
If you add back the 400 basis points on your new accounting, the biggest difference seems to be in the Microcontamination side.
Yeah, I think most of that came this year. As you saw, there was a big bump in the margin this year as compared to what we expected. As I mentioned, we got a lot of leverage out of the factories this year because that business grew quite a bit. We had made investments to be ready for the growth of the business that were dragging on us up until about a year and a half ago, and then we started to build that out. Saw a nice leverage to the bottom line with the business, combined with just success of new products and new applications and just the broadening of the business to more and more types of contamination control. We saw a big lift, and we think we continue to improve it, as I laid out in my slide.
A lot of it came this year as we grew into that capacity that we built for the business, which we really needed to do, and we did it this year.
That's great. My last question is, as I listen to the other presentations, it seems like you're not emphasizing as much on your CapEx related revenue driver. In three to five years' time, how should we think about what you used to call unit-driven revenue to be a % of your total revenue?
You're right. I think that, as we said, about 75% of what we do is unit driven, and that's going to be growing as a function of both wafer starts and the level of activity in the fabs. 25% of our business is still very much related to the industry CapEx. That part of the business probably will be growing at a slightly lower pace. We don't expect the industry CapEx to grow very significantly over the next three years. As a ratio, I would expect the unit side of our business to, on a ratio basis, to represent more going forward than what we call the CapEx products, CapEx revenues, for instance.
In our whole industry model, you didn't hear us talk about CapEx. It's A, we don't know enough about it to forecast it, and B, we just said, "We don't know. We're just going to assume it's flat with 2017." Which is already, the tool guys would tell you today that it's going to be better than that this year, but we just said, over the time horizon, we flat lined at 2017.
Thank you. Patrick Ho, Stifel. Todd, you gave a really good presentation in terms of some of the customer collaboration and the strategic partnerships you're trying to establish with your product portfolio and the technology you offer. I guess the big question is, how do you extract the right value? You do see the growth in the operating margins line, but obviously there's an aspect on the gross margin line where you can get pricing and value creation there as well. What's the fine balance there in terms of, I guess, extracting value for Entegris?
Yeah. From the partnerships. We've been really working in this mode for several years now, and I think that since the ATMI coming together with Entegris, our kind of gravitas with the main device manufacturers has grown quite a bit. They actually bring us in to talk about solutions across the divisions. Jim gave one anecdote of that in the deposition world. Actually being able to create things that no one else can create is a great opportunity for us. One, it can bring us a bigger win in terms of the revenue that we get, but also we create greater value in solving a problem that really they can't get solved another way, if we can come in there and do that kind of collaboratively.
I think, too, that the unique nature of how we can go work with our chemical manufacturers, or we can work with the wafer growers or the tool manufacturers, gives us another opportunity to actually do different things and solve the problem at the tool manufacturer sometimes at the behest of the device manufacturer or the wafer grower at the behest of the device manufacturer. Device manufacturer may have a filtration solution they really believe in all of a sudden, they really ask their supply chain to work with us on a solution at that level, when they're making raw materials for them. There's ways they can leverage that around the ecosystem through those partnerships.
Great. Maybe as a follow-up question for Jim, in terms of your discussion about the investments made on the R&D side of things and a lot of the new projects that you guys are working on today. I guess, what's the kind of life, I guess lifetime or I guess more pertinently, the R&D investments, what's kind of the life cycle of that R&D before it becomes a product, and how do you decide on these R&D projects, in terms of future product introductions?
Yeah. I think, kind of a round figure to keep in your mind in terms of sort of, I assume by lifetime you mean from inception to product. Kind of for products that go through the complete life cycle, there's a pretty wide range, but just use a figure of merit of roughly three years kind of timeframe. In terms of deciding which products, that's really the point of our comprehensive portfolio management process. Those priorities can change depending upon the success of the project, other needs that come up. It's really a very dynamic process, and it's one that we want to look at repeatedly and fine-tune it as it goes along. I think that's the business process we put in place, which really has allowed us to be efficient with our spending, but also targeted with our execution.
If I could just add to that, Patrick. If you think about the reporting that all companies do about sales from new products, then historically, you look at both at three years, which I'll actually release in the last three years. Also now more common is to look at the last five years. That's because it did take longer for the uptake to take place. You can design a product and get qualified, then that node has to ramp significantly. While you're waiting for things like 12nm, 7nm, 4nm to ramp, 10nm, 7nm nanometer to ramp, that can extend sometimes where they solve problems around the fab, then all of a sudden you have to ramp very fast, as I mentioned.
That competency to be able to do HVM on a very steep slope is something we've really taken close and said we need to have more manufacturing engineering as a percent of our R&D. Process engineering, how to actually go from a lab scale to high volume manufacturing is more of a competency that's important for Entegris. You'll get that longer term, then all of a sudden it's going to really take off, and that's kind of the shape that we see with a lot of the products now.
Hi, Mike Harrison with Seaport Global. You talked a little bit about the M&A strategy, and then you also went through a lot of the financial potential that you have on the balance sheet and the potential contribution from M&A. I'm just wondering, can you maybe give us a little bit more detail on the types of financial metrics that you look at? What do ROIC hurdle rates look like, and how do you arrive at a decision to buy versus build, or vice versa, as you're thinking about acquiring somebody versus developing a capability on your own?
We have purposely not provided quantification on some of those guidelines, I won't do that as I answer your question. I would only say that, again, we are defining very specific targets in terms of minimum levels of accretion two years after a transaction is completed, or minimum levels of ROIC three years after the deal is completed. Usually before we decide to act on any transactions, we compare the type of returns we can get from these acquisitions against just simply buying back our own stock. Usually that's one of the biggest part of the foundation of the financial discipline that Greg was describing. In terms of the second part of your question, which is when do we decide to actually develop internal capabilities versus going outside?
It's really part of this whole portfolio management committee reviews that Jim and Todd were describing in their respective presentations. I think we've become a lot better at, first of all, assessing the various options that we have in terms of where to apply and then where to put the R&D dollars. I think that we have become a lot more objective at our internal capabilities versus what may exist on the outside. PSS is a good example of that. We have internally a lot of existing sensing and control capabilities. We have concentration controller capabilities PSS had actually a better solution. They had that concentration control capability coupled with particle sizing. That combination was very powerful in the ability to provide additional process control to our customers, being able to detect agglomeration of slurries in the slurry mixture.
Again, in this particular case, we could have chosen to develop. We felt that there was a better alternative outside. We could make the financial numbers work, that was what led us to the acquisition of PSS. We're looking at options constantly.
I think Trinzik is another example of that.
Right.
That's exactly where we saw an opportunity around dilute chemistries, that we were looking at all of our opportunities that we could be pursuing, and we were placing our bets where we saw really high value and things where we could really win and we liked that portfolio. The opportunity with Trinzik gave us a chance to actually grab that segment to that niche of the market very quickly and be entrenched right away and be running right away and for the right value to the shareholder. We were pleased to be able to make that fill in the gap. We might not have otherwise done it for a while, compared to the other things we were being asked to do by the industry.
Charles Long, Goldman Sachs. I have one for Bertrand. Along similar lines, how would you view actionability on the M&A front, specifically with regards to current valuation levels in the market today and how robust the cycle has been?
Again, I think that in the back of the room, you have Corey Rucci, who is our SVP for business development. We have created a team. That team is scanning across the horizon, looking at gaps in our technology portfolio, looking at companies that could give us access to new markets, new opportunities, or new applications, even within the semiconductor space. That effort really started two years ago. I think that if I look at the quality of the M&A pipeline today, it's much greater quality, much more actionable than it was two years ago. It doesn't mean that we will be able to necessarily act anytime soon, but I hope we will be able to be a lot more active on the M&A front in the next 12 to 18 months. Valuation is always something that we take into consideration.
It's not just about finding something that is attractive and actionable, it's also we need to find something that we can afford and that we can buy at the right value. Again, as I was driving you through the guidelines that we are using, it's a mix of strategic attractiveness, actionability, but also financial attractiveness. That's the best way I could answer the question without going into specifics that we cannot go into.
I would just say, I think we have a really good process in terms of the way we do it, and again, it gets to our discipline. If you think about it, Corey's job is to find things and get deals done. He sits about 15 feet from Bertrand, and I doubt Bertrand gets his coffee most mornings before he goes and says, "What's going on, Corey?" My job, and Todd's job, is to say, "What can we do with that business when we buy it?" He owns getting the deal done. Todd owns the model, so he's got to make a commitment, and then ultimately, I own kind of the valuation and the financial metrics. There's a natural internal tension, but I think it works really well, and it keeps us all honest.
Greg, I got a follow-up. I don't think you gave out a specific free cash flow margin target. Do you have one that you can share?
I'm not really prepared to put a target out there on that today. I do think you'll see improvement over the coming years.
We probably have time for one more. Steve, you have time?
Chris Kapsch with Loop Capital. I don't want to beat this M&A one to death, but when you talked about the backdrop of the industry and how much more robust it is with the MSI growth forecast being lifted and then your ability to outperform the market being lifted and then even more positive cash flow characteristics. You kind of make the case that any potential targets looking forward could be more compelling. I'm just wondering if, given that backdrop, and given your confidence in your own outlook, if that's changed your willingness or motivation to do M&A from a metric standpoint. Conversely, you sort of made the case like, gosh, our valuation isn't really reflecting any capital allocation. There's a tension there, I guess.
At what point, absent actionability on the M&A front, do you say, "Okay, well, it doesn't look like we're going to get something done even though we'd like to do M&A, so our own valuation is too compelling.
Yeah, no. Give this team, I think, credit for having been able to manage those tensions fairly adequately. If you look at what we were able to do with ATMI, it was the result of being patient. We actually did allow cash to build up on the balance sheet, and if we had been anxious to spend that cash, we would have never been able to complete this ATMI transaction that ended up being such a value creator for our customers and for our investors. The same discipline prevails still today. What you're hearing us say is that we are going to generate a lot of cash
We'll take a measured view. We hope to be able to be active on the M&A front, but we don't know. Having said that, if in 18 months from now, we haven't been able to find the right target at the right values, you should expect us to probably return cash to shareholders in one way or another. That's really the takeaway from Greg's model, is that we're going to navigate through those options. It's very difficult, given all of the changes, factors around us, to really tell you exactly what we're going to be doing. Expect us to do something, and it's going to be one of those various options.
One last question.
I owe you one, Steve. I heard building a global infrastructure is a priority for the long term. Very simply put, could you maybe discuss talent acquisition? In other words, when you're building your infrastructure in Korea or Taiwan, in effect, I think your customers are considered national champions and are very strong purchasers of some of the same talent, or two sides of the same coin.
How do you compete? How do you feel you're doing? What might be the long-term challenge there?
That's actually a very great question. We are competing for talent, obviously, in the industry around the world. One of the things that's really helped us a lot is, really two things, is that when we got to a certain size through the acquisitions and the growth of the company, we're much better known. Within the semiconductor industry, we are actually very well-known amongst materials companies and solutions providers for contamination control. We're really a preeminent name. You walk through any fab in Asia, and you're going to see Entegris' name carrying wafers around that fab. We're not an unknown in that sense. Two of the investments that we made in local capabilities, the tech centers, really state-of-the-art tech centers. I didn't say this before, but we actually do developments with our customers in those tech centers.
They come to these tech centers to actually do evaluations of their wafers, to evaluate chemistries, integration schemes, filtration solutions. We're working side by side. We actually can source talent from some of those customers, and in many cases, we've done that. We've looked for one of the people that runs business development with the MC division and was actually a particle expert that we worked with in Samsung. Some years later, he wanted to come join Entegris. Like Wenga, he saw the writing on the wall for the importance of contamination control, and he came to be a part of the solution for that. Actually, because of our presence we have in these markets and really the gravity we have around us in the industry now, we're able to attract very good talent.
Especially if you are an expert in contamination control or advanced chemistry or physics, the stuff that we do is very leading edge. It's been able to attract increasingly eager people to want to come and work for Entegris, really in North America as well as around the world. It's always a challenge. It's a very busy market. There's low unemployment in the industry, we compete.
Okay.
I just want to add one thing, that is we've also invested in that capability. We've invested in talent acquisition. Three years ago, we didn't have a talent acquisition team. We have a North American talent acquisition team. We're building a talent acquisition team in Asia. That as well is when we talk about capabilities, that's a capability that we want to get better at.
Just a quick one, maybe for Wenga, I don't know. I think one of the trends in the industry that you serve is kind of the rise of the fabless chip maker and maybe a proliferation of more foundry-based production as opposed to integrated production. I'm just wondering whether the breadth of your product mix and maybe the breadth of your R&D capability, does it lend itself more directly to serving the foundry market longer term?
Yeah, there's a couple aspects of that question. We get great insight from the fabless companies. Increasingly, companies like Qualcomm, Nvidia, and so forth have integrators that work with their end foundry. They're hand-in-hand in defining the process with their foundry as to what will actually work with their most aggressive design. We have gone out and established relationships with the fabless companies to even get better headlights into what's coming down the pipe from their process, which are effectively the fab fillers and will ultimately consume our products. Now, we're not going to sell any specific products to them, but they are significant influencers of our end customer.
We can go in and, just back on the talent issue, we've built a core team of individuals who have worked at our customers in the past and are process integrators and really can speak that language. We can bring those individuals into the fabless companies and have a really in-depth dialogue about where their technology is going and what we would need to do to service that. That's kind of a key piece of our intelligence strategy, if you will, to kind of understand where we could play.
Yeah. I want to add a little bit regarding the foundries. If you read our financial reports, our biggest customer is a foundry customer, right? We have a lot of leverage working with them and benefiting from them to gaining some market shares. That's one aspect that we did benefit from this trend. The industry is also shifting a little bit in 2017 that the memory guy is actually getting much more power because the computation is shifting to the more memory- centric. That's why we also started benefiting a lot more from the memory makers. The other side is the remaining IDMs who refuse the foundry model are actually also growing. Those are the guys who are actually making 8-inch wafers for automotive, for IoTs.
We have revenues with them for a long time, and that actually is a growing segment for us as well. I think the foundry growth benefited us, but we're also benefiting from the memory growth and also we call the mainstream node that's controlled by the IDM.
On behalf of the team, I want to thank everyone here and on the webcast for your time today. I do want to note that we're scheduled to announce our Q1 results on April 26th, and we look forward to continuing the conversation then. This will now conclude the webcast.